CodSpeedHQ/codspeed-rust - CodSpeed

codspeed-rust

Blog Docs Changelog

Performance History

Latest Results

fix(criterion): send benchmark markers around iter_manual rounds

cod-3789-wizard-run-only-the-targeted-benchmark-cases-anchored

6 hours ago

fix(criterion): send benchmark markers around iter_manual rounds

cod-3789-wizard-run-only-the-targeted-benchmark-cases-anchored

6 hours ago

docs(cargo-codspeed): clarify BENCHNAME and --bench help The BENCHNAME help claimed a substring match on benchmark names. It is an unanchored regex forwarded to each bench binary, and the matched string differs by mode: the CodSpeed URI in simulation/memory, the framework's own name in walltime. Document that, how to use --exact, and that an empty selection still exits successfully. The --bench help now mentions it can be repeated.

cod-3789-wizard-run-only-the-targeted-benchmark-cases-anchored

6 hours ago

docs(cargo-codspeed): clarify BENCHNAME and --bench help The BENCHNAME help claimed a substring match on benchmark names. It is an unanchored regex forwarded to each bench binary, and the matched string differs by mode: the CodSpeed URI in simulation/memory, the framework's own name in walltime. Document that, how to use --exact, and that an empty selection still exits successfully. The --bench help now mentions it can be repeated.

cod-3789-wizard-run-only-the-targeted-benchmark-cases-anchored

7 hours ago

fix(criterion): send benchmark markers around iter_manual rounds

cod-3799-explore-process-per-benchmark-in-codspeed-rust

8 hours ago

fixup! feat(divan): run each benchmark in its own process

cod-3799-explore-process-per-benchmark-in-codspeed-rust

1 day ago

feat(divan): run each benchmark in its own process Adding or changing one benchmark can change the measurement of the benchmarks that run after it in the same process, because they inherit its heap and allocator state. With CODSPEED_PROCESS_ISOLATION set, divan_compat re-executes the bench binary once per benchmark. The parent keeps CLI filtering and URI generation, and passes the entry index (and the arg index for args benches) through __CODSPEED_DIVAN_BENCH_ID. The child looks up only that entry and runs it, so it allocates nothing that depends on the other benchmarks. Each child calls the instrument hooks with its own PID and gets its own callgrind dump. This relies on the runner's existing --trace-children=yes; cargo-codspeed and the runner need no changes. Analysis mode only; walltime (divan_fork) is not covered yet.

cod-3799-explore-process-per-benchmark-in-codspeed-rust

1 day ago

chore(deps): upgrade itertools to v0.15 Bump the workspace `itertools` dependency from 0.14.0 to 0.15.0, used by cargo-codspeed and codspeed-divan-compat-macros. Neither crate uses the APIs changed in 0.15 (`Position`, `all_equal_value`), so no code change is needed. itertools 0.15 keeps an MSRV of 1.63, below our 1.85. The vendored criterion fork still pins itertools 0.10, unchanged here. Closes #192 Co-Authored-By: Arthur Pastel <15641577+art049@users.noreply.github.com> Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

chore/itertools-0.15

4 days ago

Latest Branches

+1%

docs(cargo-codspeed): clarify BENCHNAME and --bench help#195

9 hours ago

d3bca8e

cod-3789-wizard-run-only-the-targeted-benchmark-cases-anchored

-7%

feat(divan): run each benchmark in its own process#194

9 hours ago

304fa86

cod-3799-explore-process-per-benchmark-in-codspeed-rust

-1%

chore(deps): upgrade itertools to v0.15#193

4 days ago

c332365

chore/itertools-0.15

© 2026 CodSpeed Technology

Home Terms Privacy Docs